Cross-modal correspondences may be semantic, statistical, structural, or affective in nature (Motoki et al., 2023; Spence, 2020a). Semantic and statistical correspondences are typically acquired through learning, either via language development or repeated exposure. In contrast, structural correspondences may be innate, emerging from the maturation of shared neural connections. For instance, the perceived familiarity of both odors and music has been linked to overlapping patterns of neural connectivity (Plailly et al., 2007), suggesting a shared multisensory neural network underlying these perceptual processes. Affective correspondences, on the other hand, arise from affective characteristics common to the sensory stimuli (e.g., perceived familiarity, intensity, arousal, or pleasantness of sensory stimuli). According to Motoki et al. (2023), these categories are not mutually exclusive and may often co-occur. In this regard, to elicit a “congruent effect” observable on behavioral or physiological measures, sensory stimuli should be selected according to their type of correspondence.

Psychological studies examining multisensory experiences often explore the effects of matched versus mismatched stimuli on affective perception (e.g., perceived pleasantness, familiarity, intensity, etc.), or on behavior, emotions, and cognition. Regarding the mutual influences of sensory stimuli, Seo and Hummel (2011) demonstrated a congruence effect between sounds and odors, likely grounded in statistical and affective correspondence. In their study, pairing a congruent sound (e.g., the crunch of potato chips) with the corresponding odor enhanced the perceived pleasantness of the odor compared to an incongruent pairing (e.g., the sound of coffee with the odor of chips). In a subsequent experiment, scientists demonstrated the existence of a halo effect, whereby the pleasantness of the sound influenced that of the odor, but not vice versa. A few years later, Seo et al. (2014) confirmed the role of olfactory-auditory correspondences in perceived pleasantness through three experiments, showing that congruent sounds not only enhanced odor pleasantness but also increased familiarity and identification. In another study conducted by Velasco et al. (2014), participants were asked to rate the pleasantness, intensity, and quality of six odors (blueberry, lemon, and orange considered pleasant, musk, dark chocolate, and smoke considered unpleasant) after listening to pleasant consonant music, unpleasant dissonant music, and white noise. While music did not directly influence olfactory perception, prior exposure to white noise reduced perceived pleasantness, sweetness, and moisture across all odors compared to the music conditions. In terms of arousal, olfactory modulation of musically induced arousal has also been observed in cases of affective correspondence (Zhou and Yamanaka, 2018).

Uni- and multisensory stimulations are also investigated for their use in psychological interventions aimed at supporting preserved behavioral, emotional, and cognitive functions. Music, for example, is well known for its benefits over anxiety and depression disorders (Aalbers et al., 2017; Gutiérrez and Camarena, 2015), as well as in regulating emotions depending on both musical and individual characteristics (Moore, 2013). Similarly, multisensory environments have also been employed for elderly individuals with neurodegenerative diseases (De Oliveira et al., 2014). In advanced stages of such pathologies, where verbal communication is severely impaired, these interventions can be adapted to focus on aiding nonverbal interaction, emotional regulation, and physiological stabilization (Ansaldo et al., 2018; Clare et al., 2020; Maseda et al., 2018). However, the mechanism through which multisensory experiences can help to maintain or enhance behavior and cognition remains under debate. One promising research direction focuses on the type of cross-modal correspondences involved. For example, Baccarani et al. (2023) study represented a significant advance in the understanding of the influence of multisensory olfactory-auditory environments on physiological recovery following cognitive stress. Following cognitive stress induced by a battery of cognitive tasks, participants were assigned to unisensory (either slow-tempo classical music or lavender essential oil diffusion), multisensory (both sensory stimuli diffused together), or neutral (neither music nor odor diffused) environments. The results highlight the effectiveness of unisensory musical and olfactory environments on some of the physiological variables measured, compared with the neutral condition. While a multisensory gain could have been expected in the multisensory condition thanks to the structural matching of stimuli thought by Baccarani et al., no beneficial effect was observed on physiological measures. If declarative data (i.e., how participants felt relaxed) could be interesting to investigate in further studies, the results could suggest that semantic or structural correspondence may not be the most relevant match. The question then arises of an affective correspondence based on the personal experience and preferences of the participants.

Cross-modal correspondence represents one of many factors that can influence behavior, emotion, and cognition. Understanding the broader impact of multisensory environments remains a complex and challenging task. A starting point could be the theoretical cognitive models, which offer useful frameworks for conceptualizing how perceived sensory input shapes cognitive processes.

Two recent articles have highlighted how a sensory stimulus (i.e., music in these articles) can influence cognitive performance by drawing on various cognitive models (Goltz and Sadakata, 2021; Gonzalez and Aiello, 2019). Before presenting these cognitive models, it is important to note that such influences can be broadly explained by the interaction of three categories of factors: the characteristics of the individuals involved, the characteristics of the task being performed, and the characteristics of the sensory stimulus.

Regarding individual characteristics, the Cognitive Capacity Hypothesis (Kahneman, 1973) provides insight into how a person may inhibit environmental information. According to this model, cognitive resources are limited and fluctuate based on the individual’s arousal level. When arousal is moderately high as opposed to moderately low, their ability to inhibit distractors is enhanced, resulting in greater cognitive capacity for task performance. Conversely, a too low arousal level may prevent individuals from processing relevant environmental information, thus interfering with task completion or behavioral adjustment. This model emphasizes the risk of cognitive overload when a distractor draws on the same cognitive resources required for the activity or the task. In other words, cognitive capacity is modulated by environmental interactions, which in turn influence performance.

Complementary to the Cognitive Capacity Hypothesis, and intersecting individual and task characteristics, the Load Theory (Lavie, 2005, 2010) proposes that the impact of a distractor on task performance depends on the type of load, perceptual or cognitive, demanded by the task. When a task involves a high perceptual load, distractors could be easily blocked out since no perceptual resources would be available and orientable towards the distractors. In this case, the sheer number of perceptual events saturates the individual’s processing capacity. However, in a task requiring high cognitive load, the current task could be interrupted by an irrelevant event similar to the activity being performed. For example, an activity that would require high visual attention capacities could be interrupted by a distractor relying on the same capacities. This model also includes the influence of aging on the capacities to inhibit distractors: a reduced perceptual capacity would attenuate sensitivity to irrelevant events, especially for tasks with low perceptual load. In other words, older adults would easily inhibit distractors during a low perceptual task due to impaired perception, compared to younger adults. This developmental perspective, which is necessary to adapt tools and spaces to people of different ages, can be found in other models or scales like the dynamic Neurocognitive Adaptation (Cieri et al., 2025; Cieri et al., 2025), which offers an interesting lifespan framework.

While Load Theory describes how distractions may impair performance, the Distraction-conflict Theory (Baron, 1986) offers a contrasting view, suggesting that distractors could exert a positive influence on cognitive performance. According to this theory, during simple or repetitive tasks with low cognitive demand, individuals could be less inclined to switch to a kind of mind-wandering thanks to a distractor (Goltz and Sadakata, 2021; Gonzalez and Aiello, 2019). Although mind-wandering has been shown to benefit important cognitive processes, such as episodic memory, empirical evidence suggests that it disrupts behavioral responses to immediate sensory external input, favoring intrinsic, self-generated thoughts (Baird et al., 2014; Poerio et al., 2017). In the case of mind-wandering during task completion, the distractor would create an attentional conflict, increasing both arousal and task-related cognitive load. People would therefore be more cognitively engaged, and the cognitive capacities for carrying out the task with the addition of the distractor would adjust to the cognitive load required for the completion of the current activity. Following this logic, in the case of a task requiring a very high cognitive load, the attentional conflict created by the distractor would add an additional cognitive load. Individuals’ cognitive abilities would be overwhelmed, making task completion impossible. In short, distractions would create attentional conflicts that would increase activation and, depending on whether the task is simple or complex, would either support or hinder its completion.

Together, these three models frame environmental sensory stimuli as either distracting or facilitating influences on cognition and behavior, depending on factors such as arousal level and task complexity. Referring to the Cognitive Capacity Hypothesis, Goltz and Sadakata (2021) explored how listening to background music affects cognitive task performance. They argued that music may interfere with the cognitive processes required for task execution, especially when it contains lyrics in a familiar language, drawing on the same cognitive systems used for reading or vocabulary learning. In such cases, music would overload the information processing capacity (e.g., via saturation of the phonological loop, see Baddeley, 1996), disrupting the ongoing linguistic task. Musical characteristics such as lyrics, musical complexity, volume, and rhythm all contribute to the interference depending on the nature of the task. Unlike music, other sensory stimuli such as smells, light, or temperature do not typically compete for the same cognitive processes. Therefore, it remains difficult to argue that these stimuli interfere with tasks via similar mechanisms.

However, according to the Cognitive Capacity Hypothesis, available cognitive capacities are modulated by arousal. Thus, other kinds of sensory stimuli could therefore indirectly influence the available capacities by modifying arousal. Odor perception, for example, is strongly linked to affective and emotional processing (Bensafi et al., 2002; Herz, 2002; Kontaris et al., 2020; Toet et al., 2020), and could impact cognitive capacity via changes in arousal. The hypothesis is inspired by the Mood and Arousal Hypothesis (Husain et al., 2002; Thompson et al., 2001), originally developed in music research but here extended to other sensory stimuli. According to this hypothesis, listening to music perceived as pleasant enhances positive mood and arousal, enhancing short-term cognitive performance. However, too pleasant music excerpts may over-activate and hamper cognitive performance, while listening to unpleasant music would have a negative impact on mood and arousal, impairing cognitive performance. Though promising, the Mood and Arousal Hypothesis has been tested in other sensory experiences, and empirical studies showed more nuanced results regarding the relationship between mood and arousal. Notably, cognitive performance can improve even in the absence of mood elevation from odor perception (Moss et al., 2008), and that trigeminal stimulation induced by certain odorant molecules could also influence cognitive performance (Lombion et al., 2009), and that unpleasant odors can enhance cognitive performance, and physiological arousal, including heart rate and skin conductance (Bensafi et al., 2002; Boesveldt et al., 2010; Brauchli et al., 1995). Such findings remind us that the relationship between mood, arousal, and sensory perceptions cannot be synthesized only by one factor, such as pleasantness.

Similarly, applying the Distraction-Conflict Theory to other sensory modalities seems challenging. For example, Ho and Spence (2005) demonstrated that peppermint odor obtained through a synthetic compound did not enhance vigilance in a simple sequential task. Instead, it improved concentration in a more complex dual-task condition. According to the authors, an increase in vigilance would have reduced reaction time without improving accuracy. However, accuracy improved only in the dual-task setting. This suggests the peppermint odor facilitated response inhibition in a complex multisensory context. It is important to note that during the dual task, the participant’s response was dependent on processing multisensory information, not just uni-sensory input. Thus, according to Ho and Spence, peppermint odorant improved the accuracy of participants’ responses by having a positive effect on the ability to inhibit dominant responses during a complex task requiring the processing of multisensory information.

To summarize, current cognitive models trying to represent the influences of sensory stimuli on cognition and behavior remain predominantly unimodal and incomplete. Although the effects of multisensory environments, such as olfactory-auditory ones, are still a matter of scientific debate, their use in multisensory designs is growing (Crisinel et al., 2013; Rey et al., 2023; Spence, 2020c, 2020d; Velasco and Obrist, 2020; Spence, 2022). Meanwhile, through the development and updating of cognitive models, it appears that the search for personalizing both uni- and multisensory experiences has been surprisingly neglected.

While traveling, cooking, walking, or engaging in sports, music accompanies us in numerous settings and supports diverse activities (Dibben and Williamson, 2007; Rentfrow, 2012). The simplified content dissemination and access to online and interactive music streaming services can partly explain the almost ubiquitous presence of music in our daily lives (Mazziotti and Ranaivoson, 2024). These platforms, operated by multinational corporations, curate information flow to optimize user engagement and profitability (Poell et al., 2019; van Dijck et al., 2019; Webster, 2023). In this manner, digital technologies supporting music streaming services have reinforced the notion of personalization, enabled by the fusion of social media, streaming services, causal inference, machine learning, and AI (Mazziotti and Ranaivoson, 2024). By suggesting music recommendations based on the ones previously listened to by the consumer, digital platforms extract and predict similarities in music taste. At a great scale, these processes are not without environmental impacts and social consequences, leading to polarization and exacerbation of class stratification (Webster, 2023; Zhuk, 2023).

When focusing on the consumer musical experience, the search for the song that will bring the most pleasure, that will help maintain a state of concentration, and physical performance are common behaviors today: out of 43,000 people surveyed by the International Federation of the Phonographic Industry (The International Federation of the Phonographic Industry, 2023), 63% of them claimed to search for specific songs in the last month, and 59% of them used personalized music lists. This trend has spurred scientific inquiry into the cognitive and behavioral effects of personalized background music. In two recent studies, Kiss et al. (2024) and Kiss and Linnell (2021) demonstrated that preferred background music (1) increases task-focused states while reducing mind-wandering ones, and (2) modulates these states (i.e., task-focusing and mind-wandering) through arousal modulation: in a context of background listening increasing arousal, the results suggested that mind-wandering states decreased, whereas task-focusing states increased. These findings highlight the influence of preferred and personalized sensory stimulation on cognition and behavior via individuals’ arousal and open the possibility of investigating such influences with other kinds of uni- or multisensory experiences.

Like music or light, other kinds of sensory stimuli can be broadcast with a click on our smartphones. Although it is still considered a curiosity or even ignored by most of the population, the number of studies investigating the possibility of broadcasting smells with small technological devices, like smartphones, is growing (Huang and Chen, 2023; Maggioni et al., 2018, 2019; Matsukura et al., 2013). Within the next few years, we will likely be able to enhance the mastering of our olfactory environment. However, the underestimated importance of the sense of smell in contemporary Western societies may delay its development (Herz and Bajec, 2022; Schifferstein, 2006; Wrzesniewski, 1999), although olfaction is a sense that develops early in humans, supporting the development of vision (Rekow and Leleu, 2023; Schaal et al., 2020). A question might arise while reading the last sentences: what is the link between personalization of our sensory environment and the development of olfactory research? Beyond being an important cultural and identity marker (Boswell, 2008; Majid, 2015), smells participate in our social life and provide important health indicators (Schwambergová et al., 2024), influence our emotional state (Villemure et al., 2003), and are at the center of sensory strategies to try to influence consumer behavior (Doucé and Janssens, 2013; Nibbe and Orth, 2017; Spence, 2015; Teller and Dennis, 2012; Yang and Cai, 2024). Odors bring personal information, and their cognitive processing is intrinsically linked to our autobiographical memory (Hackländer et al., 2019; Rey et al., 2023; Willander and Larsson, 2006). The literary anecdote of Proust’s madeleine illustrates an emotional and memory phenomenon that everyone can experience: a smell can trigger the revival of an emotion linked to a particular autobiographical memory. Numerous studies support that odor-evoked emotion can be particularly intense and that the memory recovered by olfactory stimulation could be linked to old moments in life, sometimes belonging to the first 10 years of life marketing (Chu, 2000; Herz, 2016; Jellinek, 2004; Larsson et al., 2006; Toffolo et al., 2012; Willander and Larsson, 2007). For all these reasons, the development of smell diffusion cannot neglect the concept of personalization and cannot only focus on common affective characteristics such as pleasantness, familiarity, or irritability.

The global influence of personalized sensory stimulation is still poorly understood in psychology, possibly due to the difficulty of adapting scientific protocols, but also due to the absence of a consensus regarding a method to personalize sensory stimuli. For example, studies exploring the effects of participants’ musical preferences have selected preferred music excerpt based on choices that were constrained: from a sample of musical excerpts of varying lengths, the music that induces the most pleasure or displeasure can be considered as preferred and personalized (Huang and Shih, 2011; Johansson et al., 2012; Nemati et al., 2019; Perham and Sykora, 2012). Reconsideration of musical personalization is recent and uncommon in this field of research, with participants being asked to bring their favorite CD or playlist and listen to it while performing a cognitive task (Darrow et al., 2006; Mori et al., 2014). Those studies suggest that listening to favorite, personalized music has a positive influence on certain cognitive performances, particularly attention by modifying the arousal state of the participant (Darrow et al., 2006; Kiss and Linnell, 2021; Mori et al., 2014). In other studies, the positive effect of preferred and personalized music on cognitive performance is based on factors related to the cultural background of the participants (Kotsopoulou and Hallam, 2010; Mohan and Thomas, 2020). The recent and growing interest in the use of personalized sensory stimuli for therapeutic applications (Grifoni et al., 2023), notably regarding Virtual Reality (Lee et al., 2024; Pardini et al., 2022; Pizzoli et al., 2022; Solcà et al., 2021), led researchers to investigate the importance and influences of sensory stimuli personalization. Choosing a preferred sensory stimulus over a panel or bringing one into laboratory experiments (e.g., a music excerpt) involves its evaluation and comparison. Choosing a preferred sensory stimulus over a panel or bringing it into laboratory experiments (like a music excerpt) involves its evaluation and comparison. Scientists investigate the importance of people’s metacognitive judgments on evaluations and choices regarding musical background during working, learning, or study conditions. These studies are complementary to research on individual preferences, firstly because metacognitive judgments characterize the perceived propensity of a stimulus, such as music, to help or distract a person in performing a task. Secondly, by studying the incidence of retrospective metacognitive judgments, a positive correlational link has been demonstrated between the perceived pleasantness of music and the likelihood of judging it as improving cognitive performance (Bell et al., 2023b).

In the Bell et al. (2023a) study, participants’ objective cognitive performance in serial recall tasks was impacted by the presence of music and whether the music was liked. At the same time, Bell et al. (2023a) investigated the validity of metacognitive judgments about the effects of irrelevant auditory stimuli (piano melodies and Mozart’s sonata) on cognitive task performance. The authors aimed to confirm one of two theories: (1) the direct-access account, according to which people base their metacognitive judgments on direct and conscious access to the distracting or helpful features of an auditory stimulus, and (2) the processing-fluency account, according to which people base these same metacognitive judgments on similar past experiences, whether conclusive or not. According to these two theories, the repetition of irrelevant auditory stimuli would lead to increasingly less negative metacognitive judgments. Yet, unlike the direct access explanation, fluent processing theory does not require knowledge of the precise helping or distracting characteristics of a stimulus, so an individual can make metacognitive judgments that conflict with the objectively and scientifically provable effects of a stimulus on their abilities. Through two experiments, the scientists manipulated the processing fluency of musical excerpts while maintaining their musical complexity. The scientists assumed that people are familiar and accustomed to the sound of music played in the normal direction of listening (i.e., forward), rather than in reverse (i.e., backward). The authors aimed to create an illusion of fluidity in the processing of musical stimuli and hypothesized that participants would have the illusion of performing better on a cognitive task with music played “forward” despite its complexity compared to the same music played “backward.” In other words, the scientists expected participants to wrongly judge music played “forward” as less distracting than music played “backward. Bell et al. (2023a) demonstrate that direct experience of performing a task with music helped attenuate the illusion of metacognitive judgment without eliminating it. In other words, participants modified their judgment about the distracting nature of the sensory stimuli (i.e., the background music broadcasted during the cognitive task) after the experience. Besides, participants judged the same music “backward” as being more distracting than in a normal listening mode. Finally, these results showed that human metacognitive judgments about the suitability of music to aid cognitive task performance are based on fluid processing.

This discovery highlights the potential for poor metacognitive judgments about environmental conditions that would promote good cognitive performance. In other words, we should be cautious about how a sensory stimulus, perceived as pleasant and potentially chosen among others based on metacognitive judgments, has a genuinely positive influence on our cognition and behavior. However, the lack of knowledge about the influences of personalized sensory stimuli on cognition, emotion, and behavior does not deter from the growing interest and use of AI to personalize experiences. Technological advances, such as computing power, machine learning, or data storage are combined with theoretical developments, helping researchers transform theoretical concepts into applications. The personalization of communication content and user experience is not only “based on individuals’ preferences, interests, demographics, and past behavior, item features and characteristics, or similar tastes of others, but also on psychological factors—the method of psychological targeting” (Hermann, 2022, p. 5). Individual psychological traits, such as personality traits and emotional states, are also computationally predicted by algorithms thanks to their purchase history and digital footprint (Matz et al., 2017; Matz and Netzer, 2017; Gao and Liu, 2023).

Yet, if a growing body of research gathers many fields like marketing, economy, welfare and public policy, computer sciences, and statistics, it appears that psychology could give interesting insights through theoretical models, practical feasibility, and therapeutic applications. Based on the cognitive models presented above and the research focusing on metacognitive judgments, we propose a cognitive model to represent the potential influences of uni- and multisensory stimuli on cognition and behavior.

The structure of this model (Figure 1) is based on the Cognitive Capacity Hypothesis (Kahneman, 1973), which posits that task performance relies on an individual’s limited and variable available capacity, modulated by their arousal. By integrating Load Theory (Lavie, 2005, 2010), we account for the influence of cognitive and perceptual loads from both tasks and distractors, while also considering task complexity and cognitive aging effects. Further incorporating Distraction-Conflict Theory (Baron, 1986; Gonzalez and Aiello, 2019) allows for the inclusion of potential beneficial effects of distractors on arousal. Finally, the Strength and Vulnerability Integration model (Charles, 2010), addresses age-related differences in arousal regulation. The proposed model comprises several interconnected modules (Figure 2).